Timing-dependent anti-inflammatory effects of empagliflozin in monocyte-derived macrophages from post-myocardial infarct patients with type 2 diabetes

Abstract

Background and objective

Inflammation drives early recurrent cardiovascular risk in type 2 diabetes mellitus (T2DM) patients following acute myocardial infarction (AMI), particularly within 30–90 days post-discharge. Sodium-glucose co-transporter 2 (SGLT2) inhibitors such as empagliflozin (EMPA) provide cardiometabolic benefits, but their anti-inflammatory effects and optimal timing after AMI remain unclear. Given the prognostic role of systemic markers like the neutrophil-to-lymphocyte ratio, we investigated whether early initiation of EMPA modulates NOD-like receptor protein-3 (NLRP3) inflammasome activity and inflammatory responses in monocyte-derived macrophages (MDMs) from T2DM-AMI patients.

Methods

Sixty-six participants were randomised to receive EMPA either at discharge (Arm-A) or following a 90-day delay (Arm B). Clinical data and biological samples were collected over 180 days. CD14+ MDMs and plasma were obtained at days 0, 30, and 90 (EMPA vs. no EMPA), and days 90, 120, and 180 (early vs. delayed). Inflammatory and metabolic markers were assessed using RT-qPCR, luminescence-based caspase-1 and ATP assays, and targeted immunoassays.

Results

Early EMPA administration was associated with reduced NLRP3 priming (IL1β mRNA) and activation (caspase-1 activity), potentially linked to decreased release of ATP, a danger associated molecular pattern (DAMP). In the absence of EMPA, pro-inflammatory cytokines (TNFα, IL6, MCP1) and M1 macrophage markers (e.g., CD80) either increased or remained unchanged over time. Early EMPA treatment appeared to stabilise or reduce their expression. Markers of cell senescence (p21, IL8, BCL2) were also modulated. Plasma levels of senescence-associated markers (MMP9, OPN, Serpin E1) remained largely unchanged, highlighting the importance of evaluating macrophage-specific responses.

Conclusion

Early empagliflozin administration in T2DM-AMI patients was associated with modulation of NLRP3-related inflammatory and senescence pathways in patient-derived macrophages, benefits observed when cells were stimulated ex-vivo with an inflammatory stimulus. These findings provide mechanistic insight into the timing-dependent anti-inflammatory effects of EMPA and underscore its potential for immediate post-AMI use to reduce inflammation and lower residual cardiovascular risk, supporting further clinical investigation.

Graphical abstract

Supplementary Information

The online version contains supplementary material available at 10.1186/s12933-025-03042-7.

Research Insights

What is currently known about this topic?

• Type 2 diabetes mellitus (T2DM) significantly increases the risk of cardiovascular disease (CVD), particularly after acute myocardial infarction (AMI), where residual inflammatory risk drives recurrent events.

• Sodium-glucose co-transporter-2 inhibitors (SGLT2i) provide cardiovascular protection in patients with T2DM, independent of glycaemic control.

• Preclinical studies suggest tissue-level immunomodulatory benefits from SGLT2i.

What is the key research question?

• Does the SGLT2i empagliflozin modulate the inflammatory responsiveness of monocyte-derived macrophages (MDMs) from patients with T2DM following AMI?

What is new?

• This study demonstrates that empagliflozin targets the NOD-like receptor protein-3 (NLRP3)-inflammasome, cellular senescence, and inflammation in MDMs isolated from T2DM/AMI patients.

• MDMs from patients receiving early empagliflozin therapy exhibited reduced inflammatory responsiveness ex-vivo upon lipopolysaccharide challenge.

• The findings reveal a time-dependent effect of empagliflozin on modulating innate immune memory and vascular inflammation.

Our data suggest that early initiation of empagliflozin post-AMI in T2DM patients may help modulate residual inflammatory risk, with potential implications for long-term cardiovascular benefit.

Introduction

Type 2 diabetes mellitus (T2DM) is a major risk factor for the premature onset of multiple age-related comorbidities, including cardiovascular and kidney diseases [1, 2]. Cardio-renal complications are responsible for half of diabetes-associated deaths globally, with patients experiencing an acute myocardial infarction (AMI) at a two-to-threefold increased risk of mortality compared to those without diabetes, both during and after the acute event [3, 4].

Despite established secondary prevention strategies, patients having had an AMI remain at high residual risk, with nearly a quarter of heart attack survivors experiencing major adverse cardiovascular events (MACE) that require further inpatient care 30-to-90 days following discharge [5]. Contributed to by ongoing inflammation, this residual inflammatory risk is a more robust predictor of poor cardiovascular health than dyslipidaemia and refers to chronic low-grade inflammation that persists even after the acute event has been treated [6–8]. It is contributed to by local inflammatory processes inflicting damage upon the myocardial tissue, as well as inflammatory activity affecting atherosclerotic plaque progression, the latter of which plays a key role in increasing the risk of a reoccurring event, e.g., another heart attack or stroke. Understanding mechanisms that precipitate local vascular inflammation and identifying drugs that target residual inflammatory risk, is critical to mitigating this risk and improving patient outcomes.

Cardiovascular disease (CVD) is caused by convergence of fundamental mechanisms that underlie age-related tissue dysfunction, including chronic “sterile” NOD-like receptor protein-3 (NLRP3)-induced inflammation [9], impaired cellular autophagy, and increased cellular senescence [10, 11]. Evidence supports the role of polymorphonuclear neutrophils (PMNs) and the NLRP3-inflammasome in increased vascular risk, with post-hoc analysis of the RESCUE (randomized evaluation of patients with stable angina comparing utilization of non-invasive examinations) trial reporting that inhibition of interleukin-6 (IL6) in patients with advanced chronic kidney disease (CKD), was associated with a lower neutrophil-to-lymphocyte ratio (NLR) [12].

The NLR has been identified as a risk factor for both cardiovascular and all-cause mortality [13], whilst increased inflammasome activation and cellular senescence in myeloid cells has been linked to atherosclerosis and plaque burden. These studies evidence that the number of circulating cluster of differentiation-(CD)14 + monocytes and blood borne infiltrating C–C chemokine receptor-2 positive (CCR2+ ) macrophages predict poor cardiovascular outcomes [14, 15]. Furthermore, elevated plasma NLRP3 inflammatory mediators are associated with increased risk of cardiovascular events [16], and cardiovascular and all-cause mortality, with inhibition of the NLRP3-inflammasome [17] to interleukin-1beta (IL1β) [18] or IL6 [19] pathway identified as an effective target for athero-protection. Whilst large-scale outcome studies confirm benefits of anti-inflammatory therapies targeting residual inflammatory risk in patients having had an AMI e.g., CANTOS (Canakinumab anti-inflammatory thrombosis outcome study) [18] and COLCOT (COLchicine cardiovascular outcomes trial) [20], further therapeutic approaches are required.

Originally intended as a second line agent for optimisation of T2DM, cardiovascular outcome studies have since demonstrated clinical benefit of sodium/glucose co-transporter-2 inhibitors (SGLT2i’s), over and above improved diabetes control, in patients with the condition and who are at high risk of, or have established, CVD [21–23]. These studies reveal improved diabetes care with a greater reduction in glycated haemoglobin (HbA1c) and weight loss, but more significantly illustrate meaningful reductions in MACE, hospitalisation for heart failure, and improved renal outcomes. Interestingly, SGLT2i’s confer greater protection in people with established CVD, particularly those who have had an AMI, with significant reduction in cardiovascular death and heart failure [24].

The cardiometabolic benefits of SGLT2i’s stem from their pleiotropic effects, with evidence suggesting that these compounds are immunomodulatory and provide adjunct ‘anti-inflammatory’ benefits [25]. This is supported by evidence that SGLT2i’s elicit cardiorenal benefits in the absence of diabetes, contributing to the notion that benefit is independent of glucose lowering and that they can act on cells with little or no SGLT2 expression [26]. With extensive preclinical evidence supporting the cardioprotective effects of SGLT2i’s in murine models of CVD [27–29], clinical investigations have primarily focused on systemic plasma biomarkers [30]. A post hoc analysis of the EMMY (impact of EMpagliflozin on cardiac function and biomarkers of heart failure patients with acute MYocardial infarction) trial, the first to demonstrate that empagliflozin significantly reduced N-terminal pro-B-type natriuretic peptide (NT-proBNP) levels and improved cardiac structure and function in AMI patients, found no additional reduction in systemic inflammatory biomarkers beyond that achieved with standard post-AMI care. However, whilst plasma biomarkers offer a convenient and non-invasive means of assessment, they may not fully capture the local anti-inflammatory effects of SGLT2i. In vivo animal studies have highlighted that SGLT2i’s exert potent anti-inflammatory actions at the tissue level, particularly within macrophage-enriched regions such as the infarcted myocardium and atherosclerotic plaques. Blood borne CD14 + monocyte-derived macrophages (MDMs) are key contributors to the inflammatory response post AMI. They retain a trained, immune metabolic memory, which enables a heightened or suppressed response when they re-encounter a stimulus [31] with a life expectancy in the arterial wall ranges from 5mth-to-15 yr. This heightened responsiveness is likely to be a significant driver of the vascular inflammation and damage that underpins residual inflammatory risk. Reliance on the effect of SGLT2i on systemic biomarkers may underestimate critical tissue-specific inflammatory processes that are pivotal to clinical outcomes. Furthermore, whether empagliflozin targets myeloid cell behaviour in patients with T2DM having had an AMI is unknown.

In this study we provide novel mechanistic evidence that empagliflozin targets NLRP3-inflammasome, cell senescence, and associated inflammatory events in MDMs isolated from patients with T2DM/AMI over a 6 month period. Given the heightened vulnerability to recurrent events in this population, we longitudinally profiled MDMs from patients prescribed empagliflozin prior to and 90 days post discharge, to determine if when challenged ex-vivo with lipopolysaccharide (LPS), the inflammatory response was less in those MDMs from when intervention commenced early versus delayed. By assessing how empagliflozin modulates trained myeloid inflammatory behaviour over a 6 month period, this study tests the hypothesis that the SGLT2i may mitigate residual inflammatory risk not merely through systemic effects, but by acting directly on myeloid cells that orchestrate sustained vascular inflammation after an AMI.

Methods

Ethics and approval

This was a basic science mechanistic investigation of isolated monocyte-derived macrophages (MDMs) based upon a prospective, randomised, single-blind, single-centre, double-armed study carried out between Lincoln County Hospital and the Diabetes, Metabolism, and Inflammation group (University of Lincoln) between March 2023 and January 2025. The study complied with the Declaration of Helsinki and was registered with the International Standard Randomised Controlled Trial Number (ISRCTN:12589919) registry (43113). The protocol and subsequent amendments were approved by the East of Scotland Research Ethics Service Research Ethics Committee (22/ES/0047), and the Health Research Authority (integrated research application system (IRAS:319343) [32]. This study was adopted by the National Institute for Health and Care Research (NIHR) Clinical Research Network (CRN) central portfolio management system (CPMS:54499). Participants were given prior written information with regards to the study. All participants provided written informed consent prior to enrolment.

Eligibility criteria

Individuals were eligible if aged between 18-to-84 years, with either a previously established or new diagnosis of T2DM, admitted with an AMI as per the fourth universal definition [33], who were eligible for a SGLT2i and not already receiving it at the time of recruitment. Exclusion criteria comprised patients already prescribed an SGLT2i at the time of recruitment, those planned for coronary artery bypass grafting (CABG) and those with active, ongoing malignancy or autoimmune diseases.

Randomisation and follow-up

Patients were randomly allocated to one of two arms: Arm-A (early empagliflozin therapy at day 0 of randomisation during their hospital admission with AMI) or Arm-B (delayed empagliflozin therapy at day 90 of randomisation; Fig. 1). Randomisation was performed using randomly assigned envelopes prepared by the clinical team in advance with 1:1 allocation. Empagliflozin (10 mg once daily) was commenced for patients at the appropriate times as per the protocol. Additional diabetes management was standardised and optimised as per recommendations from international guidelines [34]. Patients in both arms were followed up for 6 months with 3 follow up visits for Arm-A (days 30, 90 and 180) and 4 visits for Arm-B (days 30, 90, 120 and 180). Samples for biomolecular analysis were obtained at each visit with additional blood tests for clinical assessment performed at days 90 and 180 (Table 1). The team who performed the experiments were blinded to arm allocation for the duration of the study, whereas the patient facing clinical team, whilst aware of arm allocation, were blinded to the results from bench side analysis.

Fig. 1.

Flowchart detailing participant journey in the study and timing of initiation on empagliflozin. T2DM, Type 2 diabetes mellitus

Table 1.

Baseline characteristics of study participants

Baseline characteristics at randomisation
Characteristic	Overall (N = 55*)	Arm A (N = 281)	Arm B (N = 271)	p-value2
Age (years)	64.00 (56.09, 75.68)	61.38 (56.00, 66.27)	71.00 (57.94, 78.00)	0.083
Male sex	43/55 (78%)	26/28 (93%)	17/27 (63%)	0.007
BMI (kg/m2)	30.34 (4.95)	30.47 (5.53)	30.21 (4.35)	> 0.9
Heart rate (beats/min)	74.11 (14.68)	74.64 (13.27)	73.56 (16.26)	0.8
Systolic blood pressure (mmHg)	141.84 (25.58)	143.43 (28.78)	140.19 (22.20)	0.8
Admission to randomisation time (days)	4.00 (2.00, 7.00)	4.00 (2.00, 8.00)	4.00 (3.00, 7.00)	0.8
Length of stay (days)	4.00 (2.00, 6.00)	4.00 (2.00, 6.00)	4.00 (3.00, 6.00)	0.5
Percutaneous coronary intervention performed	47/55 (85%)	24/28 (86%)	23/27 (85%)	> 0.9
Severe left ventricular systolic dysfunction	8/55 (15%)	5/28 (18%)	3/27 (11%)	0.7
Discharge diagnosis STEMI	24/55 (44%)	13/28 (46%)	11/27 (41%)	0.7
Glucose (mmol/L)	9.10 (7.70, 13.30)	10.25 (8.20, 13.95)	8.30 (7.20, 13.10)	0.2
Diabetes				0.7
T2DM known	47/55 (85%)	23/28 (82%)	24/27 (89%)
T2DM new	8/55 (15%)	5/28 (18%)	3/27 (11%)
History of peripheral vascular disease	2/55 (3.6%)	2/28 (7.1%)	0/27 (0%)	0.5
History of atrial fibrillation	3/55 (5.5%)	1/28 (3.6%)	2/27 (7.4%)	0.6
History of chronic kidney disease	7/55 (13%)	4/28 (14%)	3/27 (11%)	> 0.9
History of heart failure	1/55 (1.8%)	0/28 (0%)	1/27 (3.7%)	0.5
History of ischaemic heart disease	11/55 (20%)	5/28 (18%)	6/27 (22%)	0.7
History of hypertension	28/55 (51%)	13/28 (46%)	15/27 (56%)	0.5
History of stroke or transient ischaemic attack	2/55 (3.6%)	2/28 (7.1%)	0/27 (0%)	0.5
ACEi/ARBs	52/55 (95%)	25/28 (89%)	27/27 (100%)	0.2
Aspirin	50/55 (91%)	27/28 (96%)	23/27 (85%)	0.2
Beta blockers	53/55 (96%)	28/28 (100%)	25/27 (93%)	0.2
DPP4i	15/55 (27%)	9/28 (32%)	6/27 (22%)	0.7
Insulin	8/55 (14.5%)	4 (14.3%)	4 (14.8%)	> 0.9
Metformin	30/55 (55%)	15/28 (54%)	15/27 (56%)	0.9
Statins	53/55 (96%)	28/28 (100%)	25/27 (93%)	0.2
Sulfonylurea	10/55 (18%)	7/28 (25%)	3/27 (11%)	0.3
Creatinine clearance (ml/min)	105.44 (71.72, 125.74)	111.27 (74.17, 128.35)	87.94 (63.76, 122.15)	0.2
Lymphocytes (× 109/L)	1.80 (1.39, 2.60)	1.72 (1.51, 2.55)	1.83 (1.38, 2.83)	0.8
Neutrophils (× 109/L)	6.91 (5.30, 9.22)	7.72 (5.76, 9.88)	6.46 (4.98, 8.94)	0.2
Monocytes (× 109/L)	0.73 (0.59, 0.91)	0.72 (0.58, 0.87)	0.77 (0.62, 0.91)	0.7
NLR	3.72 (2.68, 5.04)	3.73 (3.23, 5.90)	3.72 (2.23, 4.44)	0.2
MLR	0.38 (0.27, 0.50)	0.38 (0.26, 0.59)	0.39 (0.29, 0.48)	> 0.9
Peak troponin (ng/L)	168.00 (58.00, 709.00)	166.50 (43.50, 1, 128.50)	197.00 (88.00, 614.00)	0.7
HbA1c (mmol/mol)	55.00 (49.00, 76.00)	56.00 (51.00, 75.50)	53.00 (47.00, 79.00)	0.6
LDL (mmol/L)3	2.54 (1.13)	2.68 (1.27)	2.41 (0.98)	0.5
TG (mmol/L)3	2.00 (1.30, 3.20)	2.30 (1.50, 3.20)	1.70 (1.10, 3.30)	0.4
Total cholesterol (mmol/L)3	4.67 (1.62)	4.70 (1.54)	4.63 (1.73)	0.7
CRP (mg/L)	7.90 (2.60, 52.00)	5.30 (2.60, 98.00)	14.00 (2.30, 29.00)	0.7

¹Median (Q1,Q3); n/N (%)

²Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test; Wilcoxon rank sum exact test

³Mean (SD)

Clinical measures

At randomisation, baseline demographics, complete medical history, clinical examination, 12 lead electrocardiogram, discharge medications, and baseline transthoracic echocardiogram were recorded (Table 1). In addition, non-invasive investigations including full blood count (FBC), renal function tests (RFTs), lipid profile, non-high sensitivity C-reactive protein (CRP), peak troponin T level and glycated haemoglobin (HbA1c) were performed at baseline and then at days 90 and 180. FBC analysis was performed using XN-2000 (Sysmex, UK) and Fluorescence Flow Cytometry whilst RFTs, CRP, and lipid profile, including low density lipoprotein (LDL, mmol/l), high-density lipoprotein (HDL, mmol/l), triglycerides (TG, mmol/l) and total cholesterol (mmol/l), were performed using the c702 automated photometric analyser (Roche, UK). Troponin T was assessed using e602 automated electrochemiluminescence immunoassay (ECLIA) analyser (Roche, UK) and HbA1c using the CAPILLARYS 3TERA capillary electrophoresis analyser (Sebia, UK) as per the International Federation of Clinical Chemistry and Laboratory Medicine reference measurements (mmol/mol). In addition, a full clinical assessment, changes in medications, and occurrence of any adverse or serious adverse events were recorded at days 90 and 180.

Outcome measures

Primary

A mechanistic bench side study to determine the ability of empagliflozin (vs no empagliflozin) to blunt priming/activation of the NLRP3 inflammasome, suppress ATP release and decrease markers of inflammation, senescent cell accumulation and its senescence-associated secretory phenotype (SASP) in monocyte derived macrophages isolated from patients with type-2 diabetes with acute myocardial infarction.

Secondary

To demonstrate a mechanistic benefit and correlate the magnitude of effect observed in isolated and stimulated monocyte derived macrophages to the timing of empagliflozin therapy in AMI: early versus late therapy.

Peripheral blood mononuclear cell/monocyte isolation and macrophage differentiation

Peripheral blood mononuclear cells (PBMCs) were isolated from patient blood samples using density gradient centrifugation. Blood was layered on Ficoll Paque Plus media (Fisher Scientific, US) at a 3:1 ratio and centrifuged at 400 relative centrifugal force (RCF) for 40 min with the brakes off. Plasma was removed and stored and PBMCs were isolated from the interphase. Cells were washed twice in 10 mL phosphate buffered saline (PBS) with a 10 min centrifugation at 600RCF with the brakes on. Isolated PBMCs were resuspended and cultured in Gibco RPMI-1640 media (Fisher Scientific, US) with 10% foetal calf serum (FCS). Monocytes were isolated from PBMCs using CD14 microbeads, human (Miltenyi Biotec, Germany, Cat# 130-050-201, RRID:AB 2665482) following manufacturer’s instructions and seeded at 8 × 10⁵ cells/mL in RPMI culture media with 10% FCS and premium recombinant human Macrophage Colony Stimulating Factor (M-CSF; 20 ng/mL Miltenyi Biotec Cat# 130-096-492, CSF1_HUMAN, UniProtKB: P09603) to induce monocyte to macrophage differentiation. Cells were incubated for a total of 6 days, with media replaced 72 h after seeding using freshly prepared M-CSF (20 ng/mL). M-CSF was stored at − 20 °C in small aliquots to avoid repeated freeze–thaw cycles as per manufacturer instructions.

ATP-lite luminescence assay

Macrophages were cultured in solid, white 96-well plates in 100 µL media, with a 2 h pre-treatment of anti-ectonucleotidase ARL 67156 trisodium salt (300 µM; R&D Systems, US) followed by a 5 min stimulation ± lipopolysaccharide (LPS; 0.1 µg/mL). The ATP-lite luminescence assay system (Revvity, Cat# 6016941, Lot# 500-24065) measured extracellular ATP ([ATP]_e) following manufacturer’s instructions. Luminescence was detected using a SpectraMax iD3 plate reader (Molecular Devices, US).

Caspase Glo-1 inflammasome assay

Macrophages were cultured in solid, white 96-well plates in 50 µL media and stimulated ± Lipopolysaccharide (LPS; Ultra-Pure, Salmonella Minnesota, Merck, Cat# 43728, Lot# 3942732) (0.1 µg/mL; 4 h) and ATP (2 mM; last 1 h). Caspase inhibitor YVAD CMK (10 μM; Bio-Techne, UK) was used as a negative control (data not shown). Caspase-1 activity was quantified using the Caspase-Glo®-1 inflammasome assay (Promega, US) following manufacturer’s instructions. Luminescence was detected using a SpectraMax iD3 plate reader (Molecular Devices, US).

RNA isolation and real-time quantitative PCR

Total RNA was extracted from untreated macrophages using TRIzol reagent (Invitrogen, US), chloroform and isopropanol following the manufacturer’s instructions. A nanodrop determined RNA concentration and purity. Cellular RNA (500 ng) was converted to single-stranded complementary (c)DNA using a high-capacity cDNA reverse transcription kit (Applied Biosystems™, Fisher Scientific, Cat# 10186954, Lot# 3085877) following the manufacturer’s instructions. Samples were placed into a BioRad T100 Thermo Cycler (BioRad, US) following a 3-step protocol; (1) primer annealing—25 °C for 10 min, (2) DNA polymerisation—37 °C for 2 h, (3) enzyme denaturation—85 °C for 5 min. Real-time (RT) quantitative (q) PCR was performed using qPCRBIO SyGreen Blue Mix, ROX dye (PCR Biosystems, UK) and DNA oligo primers (Supplementary Table 1). Complementary DNA was denatured at 95 °C for 2 min, followed by 40 amplification cycles of 60 °C for 20 s and 95 °C for 5 s using a Stepone Plus Real-Time PCR instrument (Applied Biosystems, US). A standard curve gave relative cDNA concentrations, normalised against expression of housekeeping gene 18S. A melt curve analysis of 95 °C for 15 s, 60 °C for 1 min, temperature ramping by 1 °C over 20 min and 95 °C for 15 s confirmed primer specificity and detected possible contamination. Results were analysed using StepOne software v2.3 (Applied Biosystems, US).

Lumit immunoassays

Changes in secretion of inflammatory and SASP markers IL6 (Cat# 6031, Lot# 0000638525) and tumour necrosis factor alpha (TNFα; Cat# W6051, Lot# 0000641935) in LPS-stimulated cells were determined using Lumit (human) immunoassays (Promega, Madison, US) following the manufacturer’s instructions.

Analysis of plasma biomarkers

Plasma concentrations of blood biomarkers matrix metalloproteinase-9 (MMP9), osteopontin (OPN), and Serpin-E1 were measured in triplicate using SimplePlex™ Ella microfluidic cartridges (Biotechne; Minneapolis) following the manufacturer’s instructions.

Software and statistical analysis

Clinical data pertaining to patient characteristics with identifiers was documented in English and stored on encrypted servers at Lincoln County Hospital accessible only to the clinical team. Pseudonymised and encrypted data was shared with the study investigators at the University of Lincoln after un-blinding to allow for interpretation. Analysis was performed using GraphPad Prism v9.4 (GraphPad Software, USA, RRID:SCR_002798) and R statistical packages. Categorical variables were recorded in percentages (%). Normality was checked for all continuous variables using the Shapiro–Wilk test. The Wilcoxon rank sum test, Pearson’s Chi-squared test and Fisher’s exact test were used for categorical data. For intra-Arm comparisons of normally distributed laboratory results, statistical significance was calculated using a Welchs t-test. Non-normally distributed data was analysed by the two-tailed non-parametric Mann–Whitney U test. A one-way ANOVA (Kruskall Wallis or Friedman) with multiple comparison (Tukeys or Dunns) tests were used to determine statistical significance when comparing three or more groups. There was no significant interarm difference at baseline (T0) for any of the bench side parameters studied. For between group comparisons, a 2-way ANOVA or mixed effects model with Tukey’s or Šidák’s multiple comparison test was used. Data is presented as mean ± standard error of the mean (SEM). Probability (P) ≤ 0.05 was considered statistically significant.

Results

Baseline characteristics of study participants

From 361 participants screened, 295 patients were excluded due to ineligibility or not wanting to participate. Sixty-six patients met the eligibility criteria and were randomly assigned to treatment with empagliflozin. A total of 11 patients were withdrawn from the trial (6 Arm-A and 5 Arm-B), which led to 55 participants completing the study (Fig. 1). Baseline characteristics of the study participants are summarized in Table 1.

Empagliflozin reduces C-reactive protein and the neutrophil-to-lymphocyte ratio in patients with T2DM having had an AMI

C‐reactive protein (CRP) is an acute phase protein produced in the liver. It is a downstream marker of inflammation and residual inflammatory risk. Although levels correlate with the extent of cardiac injury in the acute phase of an AMI and other forms of CVD, it is not believed to have a causative role in disease [35]. Elevated CRP levels in combination with increased neutrophil-to-lymphocyte ratio (NLR) are prognostic markers of poor CV outcomes and residual inflammatory risk [36]. In T2DM/AMI patients, early EMPA (Arm-A) reduced both non-high sensitivity CRP (P = 0.0047) and NLR (P = 0.006) after only 90 days, an effect sustained over the subsequent 3mth period (Fig. 2Bi, Ci). CRP levels decreased in the absence of the drug (Arm-B T0–T90), and following delayed EMPA therapy (T90–T180; Fig. 2Bii). There was no reduction in NLR in Arm-B either without (T0–T90) or with (T90–T180) delayed EMPA (Fig. 2Cii). Lastly, we noted no significant reduction in HbA1c levels from T0–T180 in patients from Arm-A and B (Fig. S1).

Fig. 2.

The effect of empagliflozin (EMPA) on clinical parameters in patients with T2DM who have had an acute myocardial infarction. A Depicts the experimental protocol where patients were allocated to Arm-A (EMPA prescribed immediately post-acute event—blue) or Arm-B (EMPA prescribed 90 days post-acute event—purple). Clinical parameters were measured at day T0, T90 and T180. C-reactive protein (CRP; Bi Arm-A, Bii Arm-B) and neutrophil/lymphocyte ratio (NLR; Ci Arm-A, Cii Arm-B). Data represents change over time within each Arm expressed as mean ± SEM. A Kruskal Wallis with Dunns post-hoc correction was used for statistical analysis

Early empagliflozin reduced NLRP3-inflammasome priming and activation

Activation of the NLRP3 inflammasome has previously been reported to be upregulated in monocyte derived macrophages isolated from patients with CVD, including AMI and associated comorbidities e.g. T2DM as compared to healthy donor controls [37, 38]. As a major immune complex that contributes to cardiovascular injury in patients with T2DM and AMI [16, 39–41], the effect of empagliflozin on NLRP3 activity in this cohort is unknown. We sought to evaluate if this protection, i.e. reduction in inflammasome activation in MDMs challenged ex-vivo, was greater in MDMs isolated from patients where early therapy was initiated prior to discharge.

In monocyte-derived macrophages (MDMs), NLRP3-inflammasome activation necessitates IL1β priming (step1) and assembly of the NLRP3 multimeric protein complex (step2), the latter of which permits caspase-1 mediated cleavage of IL1β, a proinflammatory cytokine independently associated with the risk of mortality and recurrent MACE in patients with AMI [42]. Since IL1β can be cleaved independently of NLRP3 activity (e.g., cathepsins, collagenases, caspase-8, and neutrophil elastases), caspase-1 activation is considered the “gold standard” in assessing assembly and activation of the NLRP3 inflammasome.

Initial observations determined that there was a significant and sustained decrease in IL1β mRNA expression (NLRP3 priming) in MDMs from T2DM/AMI Arm-A patients given early EMPA (T0) at T30 (P < 0.001), T90 (P = 0.03), and T180 (P = 0.005), (Arm-A; Fig. 3Bi–iii). Inter-arm analysis identified that this benefit was also observed between our EMPA versus no EMPA (P = 0.03; Fig. 3Ciii) and EMPA versus delayed EMPA (P = 0.018; Fig. 3Diii) groups, with no effect observed in MDMs isolated from patients in Arm-B with no EMPA (Fig. 3C) or when EMPA was delayed (Fig. 3D). There was no significant inter-arm difference in IL1β mRNA expression at T0. However, IL1β expression was reduced in patient MDMs by 70% in Arm-A and 32% in Arm-B 6 months post AMI (Fig. S2B).

Fig. 3.

The effect of empagliflozin (EMPA) on inflammasome markers in monocyte-derived macrophages (MDMs) from patients with T2DM who have had an acute myocardial infarction. A Depicts the experimental protocol with patients allocated to either Arm-A (EMPA prescribed immediately post-acute event—blue) or Arm-B (EMPA prescribed 90 days post-acute event—purple). IL1β mRNA was measured in MDMs for EMPA (B), no EMPA (C), and 90 days delayed (D)-EMPA (D) therapy. ATP release was recorded from LPS-stimulated cells for EMPA (E), no EMPA (F), and delayed (D)-EMPA (G) therapy. Caspase-1 activity from LPS/ATP-stimulated MDMs (H EMPA, I no EMPA, and J delayed EMPA). Sub-panels represent T0–T30 (i) and T0–T90 (ii) within each Arm and following 90 days delayed (D)-EMPA (Arm-B) between T90–T120 (i), and T90–T180 (ii). Inter-arm comparisons between mean % change as compared to baseline; in IL1β expression, ATP release, and caspase-1 activity with EMPA (Arm-A, T0–T90) versus no EMPA (Arm-B, T0–T90); panels Ciii, Fiii, and Iiii, and EMPA (Arm-A, T0–T90) versus delayed (D)-EMPA (Arm-B, T90–T180; panels Diii, Giii, and Jiii) are shown. Persistence of effect (T0–T180) on IL1β expression (Biii), ATP release (Eiii), and caspase-1 activity (Hiii) is also presented. Data expressed as mean ± SEM. Sample size (n) is provided. For within group comparisons, an unpaired t-test with Welch’s correction analysis and two-tailed non-parametric Mann–Whitney U test were used where data was normally and non-normally distributed respectively. For between group comparisons, a mixed effects model with Tukey’s multiple comparison test was used

In infiltrating inflammatory macrophages, the NLRP3-inflammasome is activated in response to elevated levels of circulating danger-associated molecular patterns (DAMPs) e.g., ATP, which is released from injured (ischaemic) and diseased cells [43, 44]. This response is exacerbated in T2DM in the presence of hyperglycaemia, oxidative stress, and inflammation. High concentrations of extracellular ATP bind to and activate the purinergic P2X7 receptor (P2X7R) and culminate in downstream caspase-1 cleavage and maturation of IL1β. Recent studies highlight a role for the ATP/P2X7R/NLRP3 axis in multiple models of diabetes and its complications, including AMI, where P2X7R expression can predict severity of coronary artery stenosis and prognosis of AMI [45–47].

We studied the release of extracellular ATP and determined if this was lower in LPS primed MDMs isolated from patients who received early EMPA (Arm-A) versus MDMs from patients in Arm-B with no EMPA (T0–T90) or delayed EMPA (T90–T180). Intra-arm analysis determined that ATP release from MDMs significantly decreased 90 days after early EMPA treatment commenced (P = 0.005; Fig. 3Eii). Additionally, a significant inter-arm reduction was observed 30- and 90-days post treatment when intervention commenced early versus no EMPA (P = 0.04 and P = 0.007 respectively; Fig. 3Fiii) and when comparing early versus delayed in Arm-A versus Arm-B patients at 30 and 90 days respectively (P = 0.018 and P < 0.0001; Fig. 3Giii). Using a combination of LPS (priming) and ATP (activation), we also demonstrated that early EMPA decreased caspase-1 activity at both 90 days (P = 0.04; Fig. 3Hii) and 180 days (P = 0.003; Fig. 3Hiii) post therapy. In contrast, MDMs from patients with no EMPA (Fig. 3F, I) or when treatment was delayed 3 mths (Fig. 3G, J) did not exhibit a significant reduction in either ATP release or caspase-1 activity (P > 0.05). Importantly, inter-arm analysis determined persistence of effect and a significant reduction in both ATP release (P = 0.0056; Fig. S2C) and Caspase 1 activity (P = 0.027; Fig. S2D) in Arm-A versus Arm-B patient MDMs 6 months post AMI. In summary, when challenged ex-vivo with LPS or LPS/ATP respectively, inflammasome priming (IL1β) and activation (ATP and caspase-1) remained elevated in MDMs which had been isolated from T2DM/AMI patients both in the absence of EMPA therapy and when EMPA prescription was delayed. These observations support our primary outcome that when challenged ex-vivo with an inflammatory stimulus, activation of the NLRP3 inflammasome is significantly lower in (i) MDMs isolated from patients on EMPA versus no EMPA therapy and (ii) when intervention commences shortly after an AMI versus > 90 days later.

Empagliflozin reduced the expression and secretion of markers linked to residual risk and poor cardiovascular outcomes in MDMs isolated from T2DM/AMI patients

Activated by both inflammasome dependent and independent events, TNFα and IL6 are major regulators of the inflammatory response, with plasma concentrations elevated in post-MI patients at increased risk for recurrent coronary events [48, 49]. Despite the use of Canakinumab in the CANTOS trial (inhibition of IL1β), residual inflammatory risk in stable post-MI patients persisted in high-risk individuals exhibiting residual elevated IL6 [18]. With Ziltivekimab (monoclonal IL6 antibody) in trial for several cardiovascular indications, including AMI [NCT06118281] [50], we measured gene expression (TNFα) and secretion (TNFα and IL6) of these inflammatory mediators from MDMs isolated from patients on early versus delayed empagliflozin.

Early EMPA reduced TNFα mRNA expression in MDMs isolated from patients in Arm-A at T0–T90 (P = 0.06; Fig. 4Bii) and T0–T180 (P = 0.02; Fig. 4Biii) with inter-arm analysis reporting that early EMPA prevented a 113% increase in TNFα gene expression that was observed in in MDMs isolated from our no EMPA control group at 30 days post AMI (P = 0.0029; Fig. 4Ciii), with expression significantly reduced in Arm-A MDMS at T90 as compared to Arm-B (P = 0.0034; Fig. 4Ciii). Whilst no benefit was observed in our no EMPA group, delayed EMPA reduced TNFα mRNA expression at T90–T180 (P = 0.01; Fig. 4Dii). Nevertheless, it took 6 months post AMI for expression in MDMs isolated from Arm-B patients, to decrease to levels observed at just 30 days post AMI in those MDMs isolated from the early intervention group (Fig. S3B).

Fig. 4.

The effect of empagliflozin (EMPA) on TNFα and IL6 in monocyte-derived macrophages (MDMs) from patients with T2DM who have had an acute myocardial infarction. A Depicts the experimental protocol with patients allocated to either Arm-A (EMPA prescribed immediately post-acute event—blue) or Arm-B (EMPA prescribed 90 days post-acute event—purple). Expression of TNFα mRNA in MDMs was determined for EMPA (B), no EMPA (C), and 90 days delayed (D)-EMPA (D). The release of TNFα (pg/mL) from LPS-stimulated cells is shown for EMPA (E), no EMPA (F), and delayed EMPA (G). Similarly, release of IL6 for EMPA, no EMPA, and delayed EMPA is shown in panels H, I and J respectively. Sub-panels represent changes between T0–T30 (i) and T0–T90 (ii) within each Arm and following 90 days delayed EMPA (Arm-B), between T90–T120 (i), and T90–T180 (ii). Inter-arm comparisons between mean % change as compared to baseline in TNFα expression, TNFα and IL6 release with EMPA (Arm-A, T0–T90) versus no EMPA (Arm-B, T0–T90; panels Ciii, Fiii, and Iiii), and EMPA (Arm-A, T0–T90) versus delayed EMPA (Arm-B, T90–T180; panels Diii, Giii, and Jiii) are shown Persistence of effect (T0–T180) on TNFα expression (Biii), TNFα release (Eiii), and IL6 release (Hiii) is also presented. Data is expressed as mean ± SEM. Sample size (n) is provided. For within group comparisons, a two-tailed non-parametric Mann–Whitney U test was used. For between group comparisons, a mixed effects model with Tukey’s multiple comparison test was used

At the protein level, inter-arm analysis determined that early EMPA reduced TNFα secretion from MDMs in Arm-A patients as compared to MDMs in our no EMPA group (P = 0.02; Fig. 4Fiii). No significant difference was observed across our no EMPA (T0–T90; Fig. 4Fi–iii) or delayed EMPA (T90–T180; Fig. Gi–ii) group.

Early initiation of EMPA significantly reduced IL6 release in LPS-stimulated MDMs after 30 days (P = 0.05; Fig. 4Hi) and again at 6 mth (P = 0.008 Fig. 4Hiii), with inter-arm analysis reporting a significant reduction in IL6 secretion 30 days post AMI when comparing MDMs from our EMPA versus no EMPA group (P = 0.005; Fig. 4Iiii). Similarly, a greater inter-arm reduction in IL6 was observed in the first 30 days of therapy when intervention commenced early (T0–T90) versus delayed (T90–T180) (P = 0.06; Fig. 4jiii).

Although we observed a IL6 decrease in our delayed EMPA group at T90–T180, (P = 0.03; Fig. 4Jii), inter-arm analysis across the 6-month period reported that IL6 levels at T120, (30 days after delayed intervention) remained 80% higher than recorded at the time of the acute event (Fig. S3D) for Arm-B patients. In contrast, in those patients in which early EMPA was initiated, a 68% reduction in IL6 release was observed in MDMs at the 6mth period post AMI (Arm-A) as compared to baseline, versus a 28% decrease at 6 months in Arm-B patients who had received delayed EMPA for the last 90 days (P = 0.0029; Fig. S3D).

In CVD and in the context of residual inflammatory risk, activated pro-inflammatory M1-like macrophages typically initiate and sustain inflammation, whilst anti-inflammatory M2-like macrophages resolve inflammation [8, 51–53]. As shown in Fig. 4, empagliflozin impacted both expression and secretion of proinflammatory cytokines. On this basis we selected the proinflammatory markers; signal transducer and activator of transcription-1 (STAT1), CD80, and monocyte chemoattractant protein-1 (MCP1/CCL2) and assessed the impact of empagliflozin and timing of therapy on gene expression in LPS stimulated MDMs ex-vivo.

In individuals with T2DM, the majority of circulating monocytes are CCR2+ [54]. MCP1, the ligand for CCR2 is a chemoattractant and facilitator of monocyte/macrophage infiltration, effects amplified in the presence of elevated CRP [55]. Associated with increased residual inflammatory risk and MACE [56–58], recent studies report that neutralising MCP1 at the acute phase of a MI may alleviate post-MI inflammation by dampening immune cell infiltration [59]. Expression of MCP1 mRNA significantly reduced after 30 days therapy with EMPA (P = 0.002; Fig. 5Bi), with a 25% reduction 6 months post AMI (P = 0.05; Fig. 5Biii). No change in MCP1 was observed in the no EMPA or delayed EMPA therapy group (Fig. 5C, D).

Fig. 5.

The effect of empagliflozin (EMPA) on pro-inflammatory markers in monocyte-derived macrophages (MDMs) from patients with T2DM who have had an acute myocardial infarction. A Depicts the experimental protocol with patients allocated to either Arm-A (EMPA prescribed immediately post-acute event—blue) or Arm-B (EMPA prescribed 90 days post-acute event—purple). Blood samples were taken and empagliflozin prescribed at intervals indicated. RT-qPCR determined mRNA expression in MDMs for MCP1 (EMPA (B), no EMPA (C), and 90 days delayed (D)-EMPA (D)), STAT1 (EMPA (E), no EMPA (F), and 90 days delayed (D)-EMPA (G)), and CD80 (EMPA (H), no EMPA (I), and 90 days delayed (D)-EMPA (J)). Sub-panels represent changes in mRNA expression between T0–T30 (Bi, Ci, Ei, Fi, Hi, and Ii) and T0–T90 (Bii, Cii, Eii, Fii, Hii, and Iii) within each Arm and following 90 days delayed EMPA (Arm B), between T90–T120 (Di, Gi, and Ji), and T90–T180 (Dii, Gii, and Jii). Inter-arm comparisons between mean % change as compared to baseline in MCP1, STAT1, and CD80 mRNA expression with EMPA (Arm-A, T0–T90) versus no EMPA (Arm-B, T0–T90), panels Ciii, Fiii, and Iiii and EMPA (Arm-A, T0–T90) versus delayed EMPA (Arm-B, T90–T180; panels) panels Diii, Giii, and Jiii are shown. Persistence of effect (T0–T180) on MCP1 (Biii), STAT1 (Eiii), and CD80 (Hiii) mRNA expression is also presented. Data is mean ± SEM. Sample size (n) is provided. For within group comparisons, a two-tailed non-parametric Mann–Whitney U test was used. For between group comparisons, a mixed effects model with Tukey’s multiple comparison test was used

In contrast, STAT1 mRNA expression in MDMs was reduced 6 mth after EMPA therapy commenced (P = 0.03; Fig. 5Eiii). Although there was no significant STAT1 reduction at T0–T90 in MDMs isolated from our early EMPA group (Fig. 5Ei), a significant inter-arm difference in STAT1 expression was observed between our EMPA versus no EMPA group at T90 (P = 0.002; Fig. 5Fiii). In our delayed EMPA group (T90–T180), empagliflozin lowered STAT1 mRNA expression after 3 mth (P = 0.01; Fig. 5Gii) with no significant inter-arm difference observed when comparing early versus delayed response (Fig. 5Giii). Overall, STAT1 expression was 18% lower than baseline after 6 mth in Arm-B, compared to a 39% reduction for the same time interval in Arm-A where empagliflozin was initiated prior to discharge (Fig. S4C).

Lastly, whilst there was no initial benefit of empagliflozin on CD80 mRNA expression in MDMs in Arm-A across the first 90 days, a 77% difference in CD80 expression between Arm-A and Arm-B MDMs at T90 when compared to baseline control was recorded (P = 0.0015; Fig. 5liii). This translated into a significant difference observed at 6mths in Arm-A MDMs as compared to Arm-B patient MDMs where therapy was delayed (Fig. S4D). No significant reduction in CD80 expression was observed across the Arm-B time course (Fig. 5Jii).

Markers of cell senescence and its senescence-associated secretory phenotype (SASP) are reduced with early empagliflozin

Macrophage senescence is characterised by irreversible cell cycle arrest, acquisition of the proinflammatory SASP e.g., IL6, TNFα, IL8 and a compromised ability to transition from an M1-like to M2-like phenotype. In CVD, a dysfunctional immune response (NLRP3 hyperactivity) coupled with impaired autophagy leads to accumulation of senescent cells [60, 61]. Through autocrine and paracrine mediated effects, sustained SASP secretion potentiates senescent cell accumulation and contributes to a chronic inflammatory environment. Consequently, targeting macrophage cell senescence represents a potential avenue to address residual inflammatory risk, with senescent cells integral to the pathology of multiple forms of CVD, including AMI [11, 62–64].

We assessed the effect and timing of empagliflozin on the expression and secretion of markers of cell senescence and its SASP. The mRNA expression of cell cycle inhibitor p21 significantly decreased in MDMs from individuals receiving EMPA immediately post-AMI (Fig. 6Bi–iii). Inter-arm analysis determined a clear benefit of intervention with a significant difference observed between Arm-A and Arm-B patients at T30 (P = 0.026; Fig. 6Ciii) and T90 (P = 0.0013; Fig. 6Ciii). No change in p21 expression was observed in our no EMPA group at T0–T90 (Fig. 6C). However, when delayed EMPA was initiated after 90 days, p21 expression significantly decreased (Fig. 6D), representing a 35% and 23% reduction in p21 expression at 120 and 180 days after the acute event. However, a significant difference in p21 expression remained between Arm-A versus Arm-B patients 6 months post AMI (P = 0.011, Fig. S5B) highlighting the improved reduction in p21 with early EMPA therapy. In contrast, no significant reduction in the expression of cell cycle inhibitor p27 was observed, either with or without empagliflozin therapy (Fig. 6E–G).

Fig. 6.

The effect of empagliflozin (EMPA) on mRNA expression of cell cycle and apoptosis-associated markers in monocyte-derived macrophages (MDMs) isolated from patients with T2DM who have had an acute myocardial infarction. A Depicts the experimental protocol with patients allocated to either Arm-A (EMPA prescribed immediately post-acute event—blue) or Arm-B (EMPA prescribed 90 days post-acute event—purple). RT-qPCR determined mRNA expression for p21 (EMPA (B), no EMPA (C), and 90 days delayed (D)-EMPA (D)), p27 (EMPA (E), no EMPA (F), and 90 days delayed (D)-EMPA (G)), BCL2 (EMPA (H), no EMPA (I), and 90 days delayed (D)-EMPA (J)), and BAX (EMPA (K), no EMPA (L), and 90 days delayed (D)-EMPA (M)) in MDMs. Sub-panels represent changes in mRNA expression between T0–T30 (Bi, Ci, Ei, Fi, Hi, Ii, Ki, and Li) and T0–T90 (Bii, Cii, Eii, Fii, Hii, Iii, Kii, and Lii) within each Arm and following 90 days delayed EMPA (Arm-B), between T90–T120 (Di, Gi, Ji, and Mi), and T90–T180 (Dii, Gii, Jii, and Mii). Inter-arm comparisons between mean % change as compared to baseline in p21, p27, BCL2, and BAX mRNA expression over the initial 90 days period with EMPA (Arm-A, T0–T90) versus no EMPA (Arm-B, T0–T90) panels Ciii, Fiii, Iiii, and Liii, and EMPA (Arm-A, T0–T90) versus delayed EMPA (Arm-B, T90–T180) panels Diii, Giii, Jiii, and Miii are shown. The persistence of effect (T0–T180) on p21 (Biii), p27 (Eiii), BCL2 (Hiii), and BAX (Kiii) mRNA expression is also presented. Data is mean ± SEM. Sample size (n) is provided. For within group comparisons, an unpaired t-test with Welch’s correction analysis and two-tailed non-parametric Mann–Whitney U test were used for where data was normally and non-normally distributed respectively. For between group comparisons, a mixed effects model with Tukey’s multiple comparison test was used

Senescent cells evade programmed cell death, events largely regulated by the apoptosis inhibitor B-cell lymphoma-2 (BCL2), an inhibitor of autophagy, and the pro-apoptotic protein BCL-2-associated X protein (BAX). With early EMPA, BCL2 mRNA expression decreased at day 30 (P = 0.005; Fig. 6Hi) and day 180 (P < 0.001; Fig. 6Hiii) with no significant change observed in our no EMPA group (Fig. 6I). However, whilst delayed EMPA triggered a reduction in BCL2 expression at T90–T180 (P = 0.04; Fig. 6Jii), inter-arm analysis determined that 6-months post AMI, BCL2 expression was 54% lower in Arm-A, whilst only 13% lower in Arm-B when compared to baseline (Fig. S5D)—a likely consequence of the observed benefit seen in the first 90 days post AMI between Arm-A versus Arm-B (Fig. 6Iiii) coupled with increased efficacy observed in the early versus delayed response (Fig. 6Jiii). Interestingly, given the relationship between BCL2 and BAX, we observed no significant change in BAX mRNA expression in MDMs from patients in either Arm (Fig. 6K–M).

Interleukin-8 (IL8) is one of the most conserved and robust features of the SASP, with a recent study reporting that increased IL8 could predict subsequent MI or death in a 1-year follow-up of AMI patients. [65] Expression of IL8 in MDMs was significantly reduced by empagliflozin at day 30 in Arm-A (P = 0.010; Fig. 7Bi) and at day 180 (P = 0.02; Fig. 7Biii), representing a 70% decrease. While there was no significant change in IL8 mRNA expression in the no EMPA (Fig. 7Ci–ii) or delayed EMPA (Fig. 7Di–Dii) groups, there was a clear inter-arm difference of 65% in IL8 gene expression in Arm-A MDMs versus Arm-B MDMs at T30 when compared to baseline (P = 0.07; Fig. 7Diii).

Fig. 7.

The effect of empagliflozin (EMPA) on markers associated with cell senescence in monocyte-derived macrophages (MDMs) isolated from patients with T2DM who have had an acute myocardial infarction. A Depicts the experimental protocol with patients allocated to either Arm-A (EMPA prescribed immediately post-acute event—blue) or Arm-B (EMPA prescribed 90 days post-acute event—purple). RT-qPCR determined mRNA expression for IL8 mRNA in MDMs (EMPA (B), no EMPA (C), and 90 days delayed (D)-EMPA (D)), and plasma concentration (pg/mL) for MMP-9 (EMPA (E), no EMPA (F), and 90 days delayed (D)-EMPA (G)), Serpin-E1 (EMPA (H), no EMPA (I), and 90 days delayed (D)-EMPA (J)), and OPN (EMPA (K), no EMPA (L), and 90 days delayed (D)-EMPA (M)). Sub-panels represent changes in mRNA expression between T0–T30 (Bi, Ci, Ei, Fi, Hi, Ii, Ki, and Li) and T0–T90 (Bii, Cii, Eii, Fii, Hii, Iii, Kii, and Lii) within each Arm and following 90 days delayed EMPA (Arm-B), between T90–T120 (Di, Gi, Ji, and Mi), and T90–T180 (Dii, Gii, Jii, and Mii). Inter-arm comparisons between mean % change as compared to baseline in IL8, MMP-9, Serpin-E1, and OPN mRNA expression over the initial 90 days period with EMPA (Arm-A, T0–T90) versus no EMPA (Arm-B, T0–T90) panels Ciii, Fiii, Iiii, and Liii, and EMPA (Arm-A, T0–T90) versus delayed EMPA (Arm-B, T90–T180; panels) Diii, Giii, Jiii, and Miii are shown. The persistence of effect (T0–T180) on IL8 (Biii), MMP-9 (Eiii), Serpin-E1 (Hiii), and OPN (Kiii) is also presented. Data is mean ± SEM. Sample size (n) is provided. For within group comparisons, a two-tailed non-parametric Mann–Whitney U test was used. For between group comparisons, a mixed effects model with Tukey’s multiple comparison test was used

Since circulating markers of inflammation are linked to poor clinical outcomes in patients having had an AMI, leveraging the circulating SASP as an indicator of systemic senescent cell burden can potentially help identify individuals most responsive to emerging therapies and serve as surrogate endpoints in associated clinical trials. Components of the SASP can be quantified in human plasma and we studied three SASP proteins (MMP9, OPN, Serpin-E1), each of which have recently been shown to be elevated in patients post-AMI and are correlated with poor cardiovascular outcomes [66]. Whilst we observed no difference in MMP9 levels across Arm-A or Arm-B (T0–T180), a 55% difference in MMP9 levels were recorded between Arm-A and Arm-B patients 30 days post AMI (P = 0.05; Fig. 7Fiii). Similarly, whilst there was no change in plasma Serpin-E1 levels for either Arm-A or Arm-B patients across T0–T180 (Fig. 7), Serpin-E1 increased in Arm-A at T30 as compared to Arm-B patients (P = 0.0026; Fig. 7Iiii). However, notwithstanding the aforementioned observations, we did not detect any significant changes in plasma MMP9 (Fig. 7F, G), Serpin-E1 (Fig. 7I, J) or OPN (Fig. 7L, M) across the time course, neither did inter-arm analysis identify any differences between arms 6 months post AMI (Fig. S5G–I) [67] (Fig. 8).

Fig. 8.

Schematic showing the proposed molecular mechanisms through which SGLT2i empagliflozin provides cardiovascular benefit in patients with type 2 diabetes (T2DM) following an acute myocardial infarction (AMI). Individuals with type 2 diabetes mellitus (T2DM) who experience an acute myocardial infarction (AMI) remain at elevated residual risk for recurrent cardiovascular events, partly driven by persistent inflammation. This study investigated the impact of early empagliflozin (EMPA) administration versus delayed EMPA administration on inflammatory signalling pathways in patient-derived monocyte-derived macrophages (MDMs). Our data show that early EMPA treatment attenuates both priming and activation of the NLRP3 inflammasome—evidenced by reduced IL1β mRNA expression and caspase-1 activity—which is likely mediated by decreased Cx43 hemichannel activity and associated ATP release. Beyond NLRP3-dependent mechanisms, EMPA also downregulated the expression of pro-inflammatory cytokines (TNFα, IL6, MCP-1), M1 macrophage markers (CD80, STAT1), and cell cycle inhibitors linked to senescence (p21, BCL2). These inflammatory and senescent signals contribute to a self-sustaining feedforward loop that underlies residual inflammatory risk. Our findings suggest that early EMPA initiation in T2DM patients post-AMI may help break this cycle and mitigate ongoing inflammation, offering a potential strategy to reduce recurrent cardiovascular events

Discussion

As a leading cause of cardiovascular mortality, patients suffering an AMI remain at a high risk of reoccurring cardiovascular events, which persists after initial treatment, and is heightened in the presence of T2DM [3–5] with a steep increase in MACE seen particularly during the first 3 months post-discharge from hospital [24]. This residual risk is driven by increased systemic and vascular inflammation. Outcomes from clinical trials link residual inflammatory risk to the CRP/IL6/IL1 axis [18, 68] and an increased neutrophil-to-lymphocyte ratio [13, 36]. These observations are supported by preclinical and clinical evidence which link hyperglycaemia and NLRP3-inflammasome activation to increased monocyte number and enhanced myelopoiesis. Patients with T2DM having had an AMI exhibit increased plasma concentrations of pro-inflammatory cytokines including TNFα and IL6 [69, 70], whilst PBMCs and MDMs [38] from patients with T2DM demonstrate heightened expression of inflammatory mediators, e.g., NLRP3, IL1β, and IL18 [38]. Furthermore, experimental AMI accelerates atherosclerosis by recruiting myeloid cells to the atherosclerotic plaque, with post-mortem analysis of atherosclerotic lesions indicating that diabetes is associated with increased proinflammatory M1-like macrophage accumulation and necrotic core burden [71]. Drugs targeting the NLRP3-inflammasome and interrelated downstream events in innate immune cells offer multi-system benefits in cardiovascular-renal-metabolic (CVRM) conditions.

SGLT2i’s have revolutionised treatment of cardiorenal complications in patients with and without diabetes. A recent study by Kim et al. reported that 30 days treatment with empagliflozin downregulated IL1β and TNFα secretion from PBMC-derived macrophages isolated from patients with T2DM at risk of CVD. This was the first and to our knowledge, the only study to date which has reported on the ‘metabolic’ effects of an SGLT2i on patient derived immune cells [72]. However, while this work provides important initial evidence of empagliflozin’s anti-inflammatory effects in MDMs, there remains a paucity of data around the benefits of EMPA on macrophage inflammatory phenotype and associated events that underpin residual inflammatory risk in this high risk patient cohort; T2-AMI. Our mechanistic study demonstrates for the first time that PBMC-CD14+ MDMs isolated from T2DM-AMI patients prescribed empagliflozin, exhibit a significant reduction in NLRP3 priming and activation when challenged with an inflammatory stimulus ex-vivo. As evidenced by decreased IL1β mRNA expression (priming) and caspase-1 activity (activation), this reduction was observed in MDMs from patients in Arm-A at 30, 90, and 180-days post-AMI. However, an absence of EMPA or delayed therapy failed to recapitulate these observations, with IL1β levels and NLRP3 activation unchanged as compared to time of AMI, with a significant inter-arm difference observed when compared to the EMPA group. Such favourable effects are likely attributable to several downstream interwoven events. In vitro, studies have shown that EMPA downregulates TOLL-like receptor (TLR) expression, whilst Kim et al. reported that EMPA inhibits phosphorylation of the transcription factor NFκB an essential driver of IL1β priming. NLRP3 priming in MDMs is indispensable for inflammasome activation [73], the latter of which can be driven by several underlying mechanisms including lysosomal damage, increased reactive oxygen species (ROS) and changes in cytoplasmic ion concentrations [74]. Although different mechanisms are involved, activation of the NLRP3-inflammasome in both foamy and inflammatory CCR2+ macrophages contribute to atherosclerosis and poor cardiovascular outcomes. Foamy macrophages facilitate lipid accumulation and plaque growth, whilst inflammatory (non-foamy) derived macrophages, the focus of our study, amplify local inflammation and tissue injury. In sterile injury, activation of the NLRP3-inflammasome in inflammatory macrophages is largely driven by the binding of ATP to the purinergic P2X7 receptor. In AMI, high levels of DAMPs, including extracellular ATP ([ATP]_e), are released from cardiomyocytes, macrophages, and endothelial cells in response to cell injury and necrosis [75]. DAMPs are typically removed by autophagy, a highly conserved catabolic process that negatively regulates NLRP3 inflammasome activation to mitigate inflammation. However, in CVD, macrophage autophagy is impaired, and [ATP]_e increases. Experimentally, in vivo studies report that SGLT2i’s induce autophagy (via nutrient deprivation) and protect against cardiorenal injury [76, 77]. The altered pattern of differentially expressed proteins identified in proteomic analyses of blood collected in randomised trails is consistent with these findings [30]. Here, we report that MDMs isolated from patients prescribed empagliflozin immediately post-AMI, release less ATP when challenged in vitro at 90 days as compared to baseline. However, MDMs from patients in the absence of intervention, or delayed empagliflozin (T90–T180), continued to release high concentrations of [ATP]_e when challenged with LPS. Indeed, ATP release from primed MDMs in Arm-B continued to increase, with levels 6 months post-AMI only restored to approximately baseline. In contrast, the non-monotonic trajectories observed in Arm-A likely reflect the pharmacodynamic profile of the intervention. Following AMI, EMPA appears to induce an early and substantial reduction in inflammatory activity, evident at day 90, followed by partial rebound at day 180. This pattern is consistent with the re-engagement of inflammatory pathways during later phases of tissue remodelling and repair. Importantly, the initial 90 days period of reduced inflammation is clinically meaningful, as early attenuation of inflammatory burden has been associated with improved healing and reduced adverse remodelling. While convergence at day 180 may suggest waning durability, the early separation likely confers benefit during the critical recovery window. Notably, ATP release at T180 in Arm-A remained 28% below AMI levels, whereas ATP release in Arm-B MDMs persisted 14% above baseline. Therefore, it is plausible that empagliflozin blocks macrophage NLRP3-inflammasome activity through enhanced autophagy and a subsequent reduction in levels of circulating DAMPs e.g. ATP.

In the absence of pharmacological intervention, these events underscore the presence of ‘M1’ like macrophage changes, NLRP3 hyperactivation, and increased senescence, changes which underlie inflammaging and increased incidence of age-related disease [78–80]. Additionally, hyperglycaemia exacerbates the sustained polarisation of macrophages from an M2-like to M1-like pro-inflammatory phenotype, with senescent cell accumulation and its pro-inflammatory SASP linked to disease pathology in diabetes and associated cardiovascular renal metabolic complications [11, 62, 81]. Circulating levels of SASP components, e.g., IL6, TNFα, and MCP1, correlate with the occurrence of MACE, including all-cause mortality, MI, and hospitalisation for heart failure or stroke within the first year post-AMI [82]. Therefore, we assessed cytokine secretion in LPS-primed patient-derived MDMs and determined that empagliflozin significantly decreased TNFα secretion in MDMs from patients prescribed empagliflozin prior to discharge (T0–T90).

The CANTOS trial demonstrated that elevated IL6 levels, even in the context of IL1β inhibition, are strongly associated with increased risk of adverse cardiovascular events, underscoring the need for therapies that further reduce IL6 driven inflammation. In our study, empagliflozin decreased IL6 secretion by approximately 75% at 6 months post-AMI, an effect not observed in the absence of treatment, further supporting the hypothesis that empagliflozin may confer additional protection against residual inflammatory risk. Given the central relationship between IL6 and CRP, with IL6 functioning as a major upstream driver of hepatic CRP synthesis, it is notable that CRP declined in both treatment arms during the early post-AMI phase. This likely reflects natural inflammatory resolution coupled with the potent anti-inflammatory effects of high-intensity statins and guideline-directed secondary prevention therapy. The distinctive contribution of empagliflozin, therefore, does not appear to be its impact on systemic biomarkers such as CRP, but rather its capacity to reprogram monocyte-derived macrophage inflammatory memory. Persistent attenuation of NLRP3 inflammasome priming and activation, together with reduced cytokine secretion upon ex vivo challenge, indicates a cellular mechanism through which empagliflozin may provide sustained protection against residual inflammatory risk beyond that afforded by standard care. These findings raise the question of how empagliflozin may also influence upstream monocyte recruitment pathways central to post-infarct inflammation.

Newly recruited CCR2+ murine monocyte-derived macrophages amplify post-infarct inflammation and contribute to the progression of AMI injury. In contrast, resident CCR2- macrophages appear to mediate cardioprotective effects [83]. Monocyte chemoattractant protein-1 (MCP1/CCL2) is the ligand for CCR2, and a chemokine strongly implicated in the pathogenesis of atherosclerosis and AMI [84]. Recruitment of monocytes is primarily mediated by the CCL2-CCR2 axis. Blood-borne inflammatory macrophages are characterised by high levels of MCP1, and we observed a significant change in MCP1 expression in MDMs isolated from patients with early intervention. Whilst previous studies reporting that empagliflozin can lower MCP1 secretion across multiple models of CVRM disease [85–87] a potential limitation to our work is that we did not study secretion of this chemoattractant. Consequently, we further assessed two ‘classic’ markers of M1 phenotype, notably STAT1 and CD80 and subsequently measured their responsiveness to empagliflozin.

CD80 is a co-stimulatory molecule expressed on antigen-presenting cells and is a ‘classic’ marker of M1 phenotype. It plays a crucial role in the immune response in CVD and is under the transcriptional control of STAT1 which, activated by the Janus-activated kinase (JAK)/STAT pathway, exhibits increased expression in macrophages and promotes an elevated inflammatory response, facilitating IL1β secretion and increased atherosclerosis [88]. Whilst we did not observe a decrease in STAT1 or CD80 expression with empagliflozin at day 30 or 90 post-AMI, it did prevent the increase in STAT1 recorded at day 90 in MDMs isolated from patients in Arm-B (no-empagliflozin). Subsequent use of delayed empagliflozin regained expression levels comparable to baseline, i.e., STAT1 and CD80 expression were 18% and 5% lower respectively than levels recorded at the time of AMI. This compared to a 47% reduction in CD80 expression and 39% reduction in STAT1 expression observed in Arm-A 6mths post-AMI. Collectively these data infer that empagliflozin preferentially favours a less inflammatory macrophage phenotype, changes associated with a reduction in NLRP3 inflammasome activation, increased autophagy, and a reduction in cell senescence, events triggered in the presence of disease, compounded by ageing and worsened in the presence of each other. Unsurprisingly, individuals with CVD and T2DM have an accelerated ageing phenotype, caused by convergence of fundamental mechanisms that underpin age-related tissue dysfunction, including sterile NLRP3 inflammasome-induced inflammation and increased cellular senescence [89].

The SASP is a complex mixture of secreted factors that includes pro-inflammatory cytokines and chemokines. Its accumulation parallels a reduction in cardiorenal protection molecules and promotes progression and increased susceptibility of both kidney and CVD in people with diabetes [10, 11, 90]. Since SASP production empowers even a small number of senescent cells to exert widespread systemic effects, it is plausible that the SASP acts as a central mediator of cardio-renal interaction and systemic inflammation. Clearance of senescent cells using senolytics e.g., Dasatanib and Quercertin has been linked to increased lifespan and tissue repair in multiple age-associated pathologies, including kidney disease in diabetes [91]. However, more detail about safety and off-target effects of these drugs is required. Similarly blocking NLRP3 or downstream mediator IL1β (Canikanumab) raises concerns over increased susceptibility to pathogenic infection and contraindications.

Evidence suggests that cell senescence promotes the onset and progression of cardiovascular diseases including atherosclerosis, with cell senescence a recognised therapeutic target [92]. In addition to the link between cell senescence—M1 polarisation, NLRP3 inflammasome activation, and increased ATP release have also been reported to promote cellular senescence in age-related conditions where chronic inflammation is a hallmark e.g. osteoarthritis and obesity. Macrophage senescence is characterized by permanent cell cycle arrest, resistance to apoptosis, and the secretion of pro-inflammatory SASP proteins. Characterisation of the senescent phenotype is dependent upon a catalogue of morphological and phenotypic changes, including altered expression of cell cycle inhibitors, anti-apoptotic proteins, and increased expression/secretion of proinflammatory cytokines. Recent studies indicate the clinical potential of SGLT2 inhibition in preventing or delaying age-related diseases and identify the need to further elucidate the efficacy and safety of SGLT2 inhibitors as a seno-therapeutic for age-related pathologies [93].

Analysis of cellular senescence markers and SASP-associated proteins revealed that empagliflozin treatment led to a transient increase in plasma levels of SASP proteins MMP9 and Serpin‑E1 at 30 days post-acute AMI in Arm-A, compared to the no-drug control. While sustained elevation of these proteins has been linked to impaired resolution of inflammation and adverse remodelling, their early post-infarct rise may reflect compensatory, protective mechanisms. Targeted deletion of MMP9 has been shown to attenuate adverse remodelling in chronic settings, evidenced by increased apoptosis and reduced inflammatory responses [94]. However, it also facilitates tissue clearance by degrading extracellular matrix components and intracellular DAMPs, thereby enabling immune cell infiltration and phagocytosis of necrotic debris, which may limit DAMP-mediated inflammation and support initial tissue repair. The transient increase in MMP9 observed with EMPA therapy may therefore reflect a compensatory mechanism promoting these protective early remodelling processes and promoting initial tissue clearance [95]. Similarly, Serpin‑E1 (PAI‑1) is strongly induced by post-infarct inflammatory signals such as TGFβ and TNFα and can stabilize the extracellular matrix, regulate fibrinolysis, and modulate macrophage polarization toward reparative phenotypes [96]. Together, these observations suggest that the early EMPA-induced elevation of MMP9 and Serpin‑E1 may contribute to adaptive tissue repair and immune regulation, rather than representing purely deleterious effects. However, longitudinal assessment of plasma biomarkers (OPN, MMP9, and Serpin-E1) revealed no significant changes within or between treatment arms over the 6-month follow-up. Whilst biomarkers routinely inform on levels of circulating proteins in diseased conditions and enable patient stratification, local inflammation can significantly impact organ function even when systemic markers remain unchanged. It often precedes systemic involvement, operates through compartmentalised immune responses, and drives tissue-specific damage such as fibrosis or remodelling. Studying local effects enables early detection, improves sensitivity beyond systemic markers, and informs targeted therapeutic strategies. In the current study, failure to observe a response to empagliflozin across a 6mth time course could be attributable to our small sample size, as well as limitations associated with using biomarkers in a longitudinal context e.g., sample half-life, sensitivity, confounding variables [97, 98].

Contrary to these observations, early empagliflozin triggered a significant decrease in expression of cell cycle inhibitor p21 and SASP marker IL8 (chemokine), an effect not observed in Arm-B in the first 90 days post-AMI. Cell cycle inhibitor p27, unlike p21, is thought to be less sensitive to acute stress or metabolic changes, and unlike p21, is not induced by p53—a recognised target of SGLT2i [99]. Whilst p27 expression remained relatively stable in MDMs across Arm-A, there was a significant difference between p27 expression when comparing the magnitude of response between early versus delayed EMPA, a potential explanation for which may be that p27 levels at T90 for Arm-B when therapy commenced, were substantially higher than those recorded at the time of early intervention. Macrophages deficient in p21 exhibit reduced expression of SASP cytokines including IL1α, and IL1β [100], whilst increased autophagy has been shown to safeguard cardiac function, reducing p16, p21, p53, and several constituents of the SASP [101]. Furthermore, inactivation of p21 has been shown to improve plaque stability, with diminished levels linked to reduced inflammation, increased macrophage efferocytosis and apoptosis [102]. Since the ability to resist apoptosis is a hallmark senescent cells, we measured expression of anti-apoptotic BCL2 and pro-apoptotic factor Bax. As expected, empagliflozin halved BCL2 expression. However, there was no significant change in BAX mRNA expression in MDMs from patients in either Arm. With BCL2 and BAX regulated by different stimuli and with expression under the control of distinct and different transcription factors, it is plausible that this response may simply reflect variation in the pathways regulated by empagliflozin, a possibility further supported by observations that SGLT2i’s mimic nutrient deprivation, favouring augmented autophagic flux (inhibited by BCL2) as opposed to apoptosis. A limitation of this study was that we did not assess protein-level expression of senescence markers, largely due to the limited yield of primary cells available, this is noteworthy since SGLT2i’s can regulate post-transcriptional and epigenetic changes [98, 103]. Lastly, increased activity of Absent in Melanoma-2 (AIM2) and NLRP3-inflammasome have each been linked to reoccurring MACE in patients with CVD + / − T2DM [104, 105]. Whilst each activate caspase-1, AIM2 is activated by cytosolic double stranded DNA. Therefore, we cannot definitively exclude a role for the AIM2 inflammasome in baseline (unstimulated) caspase-1 activity levels detected ex-vivo. However, the expression of mRNA for AIM2 is reportedly significantly increased in polymorphonuclear leukocytes isolated from patients with AMI, but not in monocytes or lymphocytes [106].

In conclusion, this is the first study to demonstrate that CD14+ monocyte-derived macrophages isolated from patients with T2DM following AMI and treated with empagliflozin prior to hospital discharge, exhibit reduced NLRP3-inflammasome activation, decreased expression and secretion of inflammatory mediators, and lower markers of macrophage senescence when exposed to an inflammatory stimulus ex-vivo. These anti-inflammatory and senescence-suppressing effects seen in isolated MDMs were markedly diminished when empagliflozin therapy was initiated 90 days post-AMI; in such cases, MDM profiles assessed 6 months post-event (T0–T180) in Arm-B were comparable to, or worse than, those observed at baseline. Whilst many studies have explored the direct effect of exogenous SGLT2i on MDMs in an ex-vivo setting, the protective effects observed in this study likely reflect a combination of previously reported metabolic effects that underpin their protection, something that is not fully captured in an in vitro setting. However, several important considerations now warrant further investigation to determine their translational relevance and broader clinical applicability.

Limitations and future directions

Whilst our findings provide mechanistic insight into how empagliflozin may reduce residual inflammatory risk, they should be interpreted as evidence of biological plausibility rather than clinical efficacy. The study was not designed or powered to assess clinical outcomes such as MACE. Nonetheless, the cellular changes observed here align with benefits seen in large cardiovascular outcome trials of SGLT2 inhibitors and offer a potential mechanistic explanation for these effects. Future studies combining mechanistic readouts with clinical endpoints will be essential to determine translational relevance.

Several limitations should be acknowledged. First, the modest sample size limits generalisability and may explain the absence of significant changes in circulating biomarkers across the 6-month follow-up. Biomarker variability and short half-lives further complicate detection of dynamic inflammatory responses. Larger cohorts using multi-omic profiling may better capture systemic effects of empagliflozin.

Second, we assessed transcriptional but not protein-level changes in inflammatory and senescence-associated markers. Because SGLT2 inhibitors exert post-transcriptional and epigenetic effects, future work incorporating proteomics or high-dimensional flow cytometry is needed to validate functional consequences of the transcriptional shifts observed. Lastly, the proposed link between empagliflozin, enhanced autophagy, and reduced DAMP (ATP) release was inferred. Direct assays of autophagic flux, lysosomal function, hemichannel activity, mitochondrial ROS, and ATP dynamics will be required to identify these mechanisms.

It also remains unclear whether the effects observed are specific to empagliflozin, shared across the SGLT2 inhibitor class, or possibly extend to GLP-1 receptor agonists. Both classes demonstrate cardiometabolic and anti-inflammatory benefits, but their direct effects on macrophage programming and inflammasome signalling have not been systematically compared. Comparative mechanistic studies are therefore a key future direction.

Finally, the marked difference between early and delayed empagliflozin initiation suggests the existence of a potential therapeutic window after AMI. Randomised studies stratifying timing of treatment, coupled with serial immune profiling and long-term clinical follow-up, will be essential to determine whether early initiation leads to durable immune reprogramming and improved clinical outcomes.

Together, these considerations highlight the need for larger, multi-centre translational studies that integrate mechanistic immunology with imaging, biomarkers, and clinical endpoints. Understanding how SGLT2 inhibitors and potentially GLP-1 agonists or other anti-inflammatory agents, reshape macrophage biology, inflammasome activity, and cellular senescence will be crucial for advancing strategies to reduce residual inflammatory risk in high-risk cardiovascular populations.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

CEH/PES and KL conceived and designed this study. CC, MUS and JKW collected the data. CEH, CC, MUS, PES, JKW and KL drafted the paper. MI helped with the statistical analyses. All authors reviewed the manuscript. All authors critically reviewed the manuscript for important intellectual content and gave final approval for publication.

Funding

This study was funded through a European Foundation for the Study of Diabetes (EFSD) award supported by the EFSD/Boehringer Ingelheim European Research Programme on ‘Multi-System Challenges in Diabetes’ (awarded to CEH, KL, and PES) and local QR funding support (University of Lincoln—CEH and PES). All schemas and tables were produced using Biorender.

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

The protocol and subsequent amendments were approved by the East of Scotland Research Ethics Service Research Ethics Committee (22/ES/0047), and the Health Research Authority (integrated research application system (IRAS:319343). This study was adopted by the National Institute for Health and Care Research (NIHR) Clinical Research Network (CRN) central portfolio management system (CPMS:54499). Participants were given prior written information with regards to the study. All participants provided written informed consent prior to enrolment and randomisation.

Competing interests

The authors declare that they have no competing interests.